The Project Page

 Some recent additions:

Once in a while some little thing comes up that I may want to convey to one or more people. Thus, this page will be like a mailbox for odds and ends, but "facing outward". There are many little projects and tips I plan to put here. The problem is to get time for it all.

In the January -98 issue of QST⁽¹⁾ was a description by K6GWN of a phased array with two GP antennae. Gain of 6.4 dBi (about 4.3 dBd) and a turning radius of less than 40 inches are reported. Clearly, the HB9CV is smaller, has more gain and uses less material. It has a much smaller back lobe as well. Stack two HB9CV's instead!

It is not my intention to come down on -GWN. He did exactly the right thing. Experimenting is an important part of amateur radio. Many great innovations has come from this. And some not so great.

Let me tell you about the background of the antenna described here, as I know it:

The antenna was supposedly invented by HB9CV for 14 MHz and scaled by someone else to 145 MHz.

In the first years of 1970, while I still lived in Sweden (as SM6CVV), the HB9CV became quite popular. My stacked pair was sometimes referred to as "HB9CVV"! Drawings circulated and there were as many theories about the way to match them as there were hams building the antenna! I tried them all, and none gave more than a half way descent match. Some time later I invented a method of plotting the reflection coefficients from any load on a Smith chart. Once I saw the nature of the impedance in the feed point, (as I remember S₁₁ was about 0.6 at -120°) I could device at least two solutions. One high pass and one low pass. I choose the low pass solution, described here below.

The boom and elements are made of brass tubing, 8 and 4 mm diameter. The driving lines are 2 mm brass wire, spaced by 2 mm from boom and elements. The dimensions are hard to read here, but the front and rear elements are 952 and 1031 mm long respective.

Spacing is 251 mm. The feed is 163 mm out from the boom center on the front element and 166 mm out on the rear. 108 mm from the front element it crosses over (or through with isolation) the boom. The feed point is in the corner near the boom and the upper element. So, the 22 pF capacitor is connected between the boom and the corner of the feed line, and the inductor is in series between this point and the coax. Adjusting the coil and the capacitor will give a just about perfect load.

The antenna provides 6.2 dBd gain with a 75° angle in the E-plane. As measured at the 1992 VHF Convention in Ventura, CA. It won the class for 2 element 2-meter antennae.

I have a feeling that this antenna could be even better, but not by much, and I would be interested to hear from someone willing to take it on with an electromagnetic simulator. In a few months I will hopefully be able to do so myself. Eagleware is coming out with an EM simulator.

Possibly pure 50 W match can be had somewhere along the feed line if the points where they are attached to the elements are moved? Or, can a capacitor to ground somewhere from the line provide match somewhere else along the same? F/B ratio and gain are probably already about as good as they can be with the approximate dimensions given.

1) QST, January 1998, pp 61 - 63. "A 2-Meter Phased-Array Antenna", Harold "Hal" Thomas, K6GWN.

My HB9CV ready to receive 137 MHz weather satellite signals. OK, it is out of band for this antenna, but it works!

Power meters are always fairly costly items, even on flea markets. It is because they are so useful. Power meters and flea markets both actually.

For this project you will need a cable with crimped BNC connectors. The cable will soon break at the crimps. They are notorious for this. Salvage the connectors and the center pins! Get a feed-through capacitor and a diode, 1N5711 or 1N34 (in the small glass package). In the picture below you can see two assembled probes, one probe connected to a 50 Ohm 5W load, one connected to a BNC-T for monitoring a signal going through, a feed-through capacitor, a diode and a 2W load. The diode "sniffs off" the RF voltage without loading the signal.

With loads of the kinds pictured here, where there is a connection right through, the probe can of course be plugged right to the other side without a T-connector.

Now, cut a piece of paper to cover the dial of your favorite multi meter, such that you can see the outermost part of the needle. Calculate what voltages would result from various power levels.

V=Ö 2•P•R

That is: the square root of the result of: 2 times Power times Resistance. R is usually 50 W . Notice the 2 coefficient! The diode probe will read the peak value of what hopefully is a sine wave. This peak is 41% higher than the average value. 1.414… multiplied by itself is 2. This dial assumes a 50 W load.

On the other side of the paper I have a 75 mW dial for the 2.5 V range of this particular instrument.

For such low voltages as a few Volt (the deflection may be 1/5 of full scale, 0.5V) the voltage drop in the diode must be considered! For the Shottky 1N5711 it is about 0.3 V for the very small currents used here. For 1N34, a Germanium diode that the bit pushers these days has decided is outdated, it is about 50 mV, maybe less. (Subtract from the calculated meter reading before putting the marks on your overlay!) 1N34 is an excellent small signal detector diode, but I guess nowadays it is GaAs, Si or "Wireless" (a buzzword for radio) or it is not at all. I have even seen "Wireless Antennas" mentioned. I guess this eliminates consideration of Yagis, the little cellular phone show off stick on the side window of the Suburban (Oh, no! Not that!) long wire dipoles and ground plane antennae, leaving patch antennae, parabolic reflectors and wave guide horns…

The 1N5711 can handle some 70 V if I remember right, and I do not think the 1N34 is far behind. It is worth remembering that when the meter reads 50 V, the peak value of the other half period is also 50 V but in the other direction. So, the diode is subjected to a reverse voltage of 100 V. In other words, power levels above 10 - 15 W may pop the diode. It may be wiser to use an attenuator ahead of the load with detector attached to the latter.

In the picture above a 20 W and a 30 W attenuator with N-connectors and a 2 W attenuator with BNC connectors. In another part, another day, later on this page, I will describe how to build pretty good loads and attenuators with regular resistors, using a method I came up with years ago. Attenuators with 15 dB or more attenuation are good loads. The signal going through a 20 dB attenuator is attenuated by 20 dB (100 X in power), totally reflected at the other end, be it short, open or whatever, reflected back through, attenuated by another 20 dB before reappearing on the input. This way, a load with a Return Loss of 40 dB is had. Meaning that only 1/10.000 of the incident power comes back. A very good load indeed. In reality the value of the 50 W resistance in the attenuator is probably not accurate enough to give such a near perfect load.

The 1N34 diode works up to about 1.5 GHz, and the 1N5711 to 1.2 GHz or so. It is mostly the diode package that limits the upper end. The load at these frequencies must of course be good as well. It is questionable if one should use a BNC-Tee to connect the detector above 500 MHz.

The diode itself presents very little load, unless one draw a lot of current from the DC side. I will check how much harmonics it generates and present the result later. (January 3, 1998: I checked with my IC-2AT on a spectrum analyzer. 5 W power level. It had a second tone, 290 MHz, at -20dBc and the rest at least 60 dB down. This did not change at all when I added the diode probe, even with a load on the DC side.) The T-form, with one side open stub will of course introduce a mismatch. Here below, a model run of this configuration in =SuperStar=:

S21 (blue-green) is the Insertion Loss in dB, V11 (red) is the VSWR from 1:1 to 3:1 and S11 (dark blue in the Smith Chart) is the Return Loss (another measure of match) in dB.

We can see that with a stub of 30 mm length and a diode capacitance of 0.2 pF the VSWR at 430 MHz (markers 3 in the left graph and 7 in the right) is 1.59:1 from this device. Corresponding to a return loss of 12.86 dB. Borderline acceptable. The resulting insertion loss, from reflected power, is 0.23 dB. We can also see that it is quite good at VHF. A 50 W cable (and connectors) has about 1 pF/cm and the diodes' 0.2 pF is a guess so it would not matter much (to the matching) even if the diode had 1 pF.

Recently a visit to the flea market paid off. Again! A beautiful AN/PRM-1A with a complete set of antennae, cables, probes, adapters, external meter, manual and line power supply. All in shape like new.

It is a battery tube radio. They require a 1.5 V for the filaments and two 90 V batteries (in parallel) for the plate voltage. The 90 V batteries are expensive, hard to get, do not last long and have a poor shelf life. Well, the radio can be operated on the 115/230 V AC power supply that came with it, and it worked fine, but it is not portable. I decided to build in a DC/DC converter. The components I had in mind run at about 100 kHz. For a receiver covering 150 kHz to 25 MHz this is a daring step! Like putting a transmitter inside the case and hope not to hear it! Well, let us see what de-coupling and shielding can do!

My first version of the Switch Mode Power Supply. An LT1074 to the left switches down the 12 V to 1.5 V and 0.8A for the filaments. To the right: an LT1162 switches up 12 V to 75 V / 50 mA. (90V was not needed.) This version, after adding a few shields, worked reasonably well and did not make too much RFI noise. The Linear Technology switcher ICs' are surely not the only ones on the market, but they have worked very well for me.

The right side of the receiver, front panel up. At the bottom: a new back plate. I did not want to drill even small holes in the original one! A 7 Ah/12 V gel battery on the lower left (gray and black) and the converter mounted on the back plate to the right of the battery, under "M" in MAR. Plenty of room left over!

Behind the lower right corner of the front panel, as seen from the outside: the added switch for battery operation. I needed one more switch built in, disconnecting the converters from the battery when the radio is turned off. A cord around the switch shaft, these linen phenolic details and a micro switch do the job.

This was a typical job made more difficult because of the vocational school background! "Mr. Do-It-Your-Self" would have blasted a .45 caliber hole anywhere in the front panel and put an extra switch in. The solution I came up with required no alteration of the original cables and no drilling or filing of anything in the radio. Just the one screw holding the linen phenolic board is exchanged for a longer one. The cord to the shaft of the switch runs through an opening between the decks of a function switch and a lot of cables. I think my old teacher, Ivar Nilsson, would have approved of this modification. If so, it is done right. Funny how things stick with you! 37 years later. He must have rubbed it in real hard.

On the upper right of the front panel you can see the antenna input switch positions ANT - LOOP - CAL. In the CAL position there is no external connection to the receiver input. I can then hear the two inverters slightly apart in frequency. At 200 kHz they give about a 50 µV reading, then less and less for the higher harmonics. Above 1 MHz it is barely readable, but all the way up to 25 MHz they can be heard as a slight change of background noise. With an antenna connected they can be heard over the space noise only < 1 MHz. Later I re-built the converter, improving the layout a little, and with some more ferrite chokes and Tantalum capacitors for better de-coupling. It got better. To get this converter all quiet it probably has to be built into a tinned steel can with filter compartments for everything that goes in and out. The schematic is straight from the Linear Technology Data book. I have had very good experiences with their switch mode controller IC's, and this project was no exception. They are easy to work with and very well behaved. The regulation of the output is excellent. No need to adjust the A+ and B+ from the front panel anymore. These are the voltages of the filament and the plate batteries. Since they drift all the time, so one has to check and adjust the voltage frequently for accurate measurements. These inverters draw about ½ A with the radio on and 45 mA, from the 12 V battery, with the radio disconnected as a load. The latter is too much to leave on, since it would discharge the battery in about a week. Therefore the need for the extra switch in the bottom picture above.

Listening to short wave is an interesting hobby. "DX-ing" as it is called. The Short Wave Listener is often referred to as an SWL, already back when cryptic acronyms were few and far between.

Well, why a loop antenna? "The wire out the window to the tree in the backyard does just fine! I get zillions of stations on that one!" Depending on what receiver you have, that may just be the problem! Unless quite sophisticated, and quite expensive, most transistor radios of today do not have much, if any, pre-selection. That is: everything on the antenna enters the mixer in the radio. Mixing is what it does! One station with another, and all the resulting signals and their harmonics, as well as the original ones and their harmonics, are mixed with the local oscillator, and its harmonics, in the radio. The result is a "forest" of stations that does not even exist! Connecting a long wire antenna to a spectrum analyzer is a harrowing experience. There are signals everywhere. Some have a level of many mV (1/1000 Volt). Imagine each signal as a rotating arrow! It rotates with the speed of the signal frequency. For example 9,590,000 times per second for BBC at 9.59 MHz. Now let each frequency be represented by such an arrow, the length of which is its signal strength. Line them up, point of one to the tail of the next. Let them rotate at their frequencies. It is not hard to imagine how they mostly cancel each other, since the frequencies are not synchronized to each other. Just for this reason several of them will now and then, momentarily, add in the same direction. Say 100 signals of about 5 mV adds up. It is 0.5 Volt! Far more than the receiver can handle in a clean way. You have intermodulation!

A pre-selector will narrow the range of frequencies reaching the receiver. It will not help the case where one or several very strong signals are right next to the one you are listening to, but it helps for more removed frequencies. The Loop Antenna to be described here has a Q of about 80 at 10 MHz. This means that outside a band of 10/80 MHz = 125 kHz other signals are down by 3 dB and of course even more further out.

The Loop Antenna.

The loop in this antenna is made out of a 2 m long, 8 mm thick, aluminum tube. As you can see, it is bent to form a loop. It may just as well be round (but then it did not fit in my suit case). It does not have to be a 2 m long tube, or even a tube, either. I built one version with a Litz wire around the inside edge of my briefcase, and it worked fine. I enjoyed, somewhat, listening to Radio Sweden International from a hotel room in Southern China!

On the left above is the schematic diagram. Simple enough! The variable capacitor is carefully selected for the project as the first one I saw in my "Box with Noble Junque" with a reasonable size and capacitance. The coil was determined experimentally, but I have later modeled the antenna and found it to be 120 nH for this size of loop. A smaller loop will need less coil. You can see it in the "Even closer view" above. This is the 2-turn coil you can see on the right side in the right picture. The switch in the middle and the smaller variable capacitor do not work very well. The larger capacitor, 360 pF, works fine, and extra 360, 750 and 1500 pF can be switched in with the DIL switch.

On the right above you can see what goes on in the feed point as the capacitor is tuned, using different coils. Too small, too large and just right. The plotted circles in the Smith Chart are the result of varying the capacitor. The different circles are the result of three different coil values. Clockwise end of each circle is for the antenna tuned to too low a frequency and the center of the Smith Chart is the tune for a perfect match. As you see, the tuning of the capacitor and the size of the coil result in perpendicular moves, so a perfect match (for one frequency at a time) is truly possible. An interesting property is the fact that the match stays excellent over the tuned range, from a few MHz to above 15 MHz.

This may not be so important, I bet your receiver is not a perfect 50 W load anyway!

Even though this is not a "clean" and nicely symmetrical loop antenna it works great. The tuning capacitor should really be in the middle, on the top of the loop, and the middle of the bottom should be grounded, but this is not a practical solution. One nice feature of loop antennae is the ability to "null out" signals that are not desired. By rotating and/or tilting the antenna. Power line buzz for one. Another feature is that loop antennae are "probes in the magnetic field". All antennae that are considerably smaller than the wave length (at their operating frequency) are probes. The 5 m of long wire is a probe in the electric field of a 300 m (1 MHz) wave. The diameter of this loop antenna is only 1/50 of the wavelength of a 10 MHz signal.

As a TEM wave (a regular radio wave travelling in free space) comes in close to buildings and ground some of it reflects. The electric fields partially cancel each other in this area. The magnetic fields partially add. Since we are ground and house dwelling creatures, this suits us and the magnetic probe fine!

Save yourself the effort of trying to extend the frequency range upwards by much! I put in a switch and a smaller variable capacitor for possible higher frequency operation. It did not work. The larger capacitor, ~360 pF, allows for operation to about 22 MHz and it is as high as it gets. The 2 m rod in itself will become l/4 long at 37.5 MHz, and it is the absolute limit for any fundamental operation of this antenna. Stray capacitance of some 10 pF and the minimum capacitance of the variable lowers this to about 22 MHz.

Lowering the low end is possible. For this purpose I added a little, pretty poor performance, DIP switch. The red thing with four white pins on. With the first I add in an extra 360 pF, the next adds 750 pF, next 1500 pF and the last 3300 pF. This way they can be added as in a binary code for 16 possible combination (including all out) The two last switches give no practical use. They result in very narrow tuning ranges at a few MHz. The low end, without switching in extra capacitors, is about 6 MHz, which is good enough. The last one or two switches are almost meaningless since they only add a MHz or so.

Now, go build your antenna and let me know how it worked for you!

Many years ago, while one still could get film for movie cameras, I decided to build a timer that could handle any of the five (5!) film cameras I have. The trigger is in different places, so the mechanism should be adjustable. I also wanted a wide range of times. The film is played back with 16 frames/second for Double-8 and 18 fps for Super-8. 16 mm can be played with either, or even 24 fps. If one takes 1 frame/minute the event is speeded up by 960 x on D-8. This DL-8 camera is spring wound and takes about 450 frames on one winding, 28 seconds worth of play. This is usually plenty for one scene.

The timer, front view…… ……………..with a Bolex Paillard DL-8 mounted

The left knob sets the speed of a slow RC oscillator, running from about 1 to 0.5 Hz. The right knob is a switch, selecting divider in a series of ÷2 circuits. The CD4040 is an inexpensive CMOS IC containing 12 dividers in cascade. Maximum division ratio is then 4096 and if the oscillator is set to 2, there will be 8192 seconds or 2 hours 16 minutes between exposures. Set the multiplier to ~1.2 and the switch to 256 for 300 s or 5 minutes! Another click on the switch: 10 minutes. One more click: 20 minutes. A very handy device.

The final output triggers a SCR that starts a geared motor. After a short run, a micro switch shunts out the SCR, turning it off, and after one full turn, when the micro switch opens again, the motor stops, awaiting a new trigger signal! With an ON-OFF-ON switch on the front I can select if the arm stops at the left or the right end. Some cameras can be set to stay open for the entire delay time, then switch frame and stay open again. This way one can film the stars of the night sky moving! In the normal mode and good light, the exposure time is determined by the settings on the camera. The current consumption from the CMOS circuits is microscopic, making the motor just about the only load on the battery. The mid position of the switch disconnects the battery. I use a small 9V battery and a very high grade Escap motor with an iron-less rotor, geared down about 300 times. This makes the movement of the arm a bit slow, so I cannot use the timer for less than about 3~4 s, the time it takes for the arm to make one cycle.

Do not leave unused inputs hanging! A CMOS can die from this! Use the remaining 4 inverters in the 4069 to drive the SCR gate! Circuit works fine from ~3V to 15 V supply. You may have to let the motor determine this.

I made films of snails in the yard outside the house. I filmed seeds growing and flowers unfolding. I filmed clouds, I filmed the midnight Sun for 24 hours, barely setting at Midsummer in Umeå, just below the Polar Circle. I filmed candles burning down, water moving from one vessel to another via a wick. I have got many miles out of this timer!

Good luck!

For years I have been thinking about building a delay circuit for quite fast events. The primary use would be to repeat the famous picture Edward Edgerton took of a milk drop hitting a plate covered with milk, splitting in a crown of small drops. Or to freeze a bullet in flight. Obviously one can detect an interrupted light beam, a sound or any other signal, and a little bit later, with good accuracy, set off a flash, exposing the film in a camera with an open lens in a dark room. For a truly versatile timer one will have to handle intervals from µs up to seconds.

Here is one (under construction):

The ideal CMOS IC from the "4000-series" for this project is the CD4538. It has two independent timers, but we need only one, so on the unused side the T1, A, and CD inputs are to be grounded, B to V+ and T2 to be left open. One of the great advantages of this timer, as opposed to the 555, is that the time = MW x µF in seconds! Simple and straight forward. The R should not be less than ~5 kW . I used 4.7 kW . The variable R (a potentiometer) should thus be considerably more, for a useful range. I use a 10 turns 100 kW pot with a dial. I adjusted the dial to show 0.47 when the potentiometer is at 0 Ohm. So I get a range of 0.47 to 10.47 with a resolution of 0.01 corresponding to a resolution of 100 ns for the fastest range of 4.70 µs to 104.7 µs and 10 ms on the slowest range of 0.47 to 10.47 s. Switching in 6 different capacitors, 10 nF to 100 µF sets the ranges.

For a truly general purpose device, the input signal must be conditioned. The A input triggers on positive and the B input on negative going ramps. (Trig level is about 50% of the supply voltage.) It is a great advantage to have both polarity triggers, but a pre-amplifier with variable gain and output level will allow for setting the trigger to the best edge from the source. The outputs are "complementary", meaning that when one goes high, the other goes low. It is also very useful for general purpose applications. They cannot drive much current, or stand much voltage, so a couple of transistors or, nowadays, power MOS FETs can be very useful. They may have to stand 150 V for to trigger a flash. A small SCR, rated 600 V may be another good idea. They trigger flashes very well.

So far I have not finished these input and output details, but the timer part works nicely. Some kind of connector supplying a probe with a LED and a photo transistor, would also be a good idea. A 5-pin 180° DIN audio connector is probably a good choice. Ready made, shielded, cables can be had in a Hi-Fi store for a few $$. Just avoid the gold-plated, oxygen free scams!

 I have a couple of stroboscopes that give a very short duration flash. You can actually use a regular camera flash if it has either an automatic exposure setting that "bites off" the flash if there is enough light coming back, or if you can set power fractions on it, like 1/64 or so of full power. The flash is then very short, maybe on the order of 50 µs. Test it on something fast, like a grinding machine, and see if the details on the wheel are "frozen!" You do not have to use film to see this, it is clear to the naked eye.

A regular flash, without automatic exposure circuits, may have a duration of 1/1000 s or even longer, and a lot of afterglow. It will work too, but not on really fast moving objects.

Now to something much more complicated, but useful! Sorry for the 190 k image, but I have compressed it!

When building RF amplifiers, usually with transistors, they do not have 50 Ohm impedance in or out. Yet, 50 Ohm is usually what we want to match them to. For best noise performance, the input may have to be off a little.

Below is a chart I have calculated for 860 MHz, the cellular phone band. I had to put it somewhere! It can be scaled to any frequency. For 3 times 860 MHz, the dimensions become 1/3. They are in inches here, 25.4 mm. The thickness of the board and width of the lines do not scale with frequency, just the lengths of the L₁ and L₂

Look at the top of the diagram! There is something that looks like an upside down 'T'! It is the matching network. It is made of copper foil, or circuit board laminate. The lines are 0.0565" (1.44 mm) wide and the board material is regular FR-4 (glass fiber) 0.031" (0.79 mm) thick.

On the left side is the device to be matched, the output of a transistor maybe. On the right side is a 50 Ohm load that the transistor will be matched to.. Say the transistor has an output impedance of 100 Ohm! In the upper diagram there is a '100' line looping down. Its lowest part is for L₁ = 2.25", but looking at the bottom graph we can see that at about 2.30" it will cut the '0'- line. (No inductance or capacitance to be added, resistive match only.) From both diagrams we can see that the stub L₂ should be some 0.8" long.

If the transistor also have, say, 50 Ohm of capacitance, "Negative Reactance" we have to cancel it with equally much "Positive Reactance." On the lower chart we can see the '50' line (not the '-50') to the left of 2.3" It looks like the 100 Ohm and +j50 cross at about L₁ = 2.2" and L₂ = 1.0". With these values the transistor should be matched to 50 Ohm.

Impedance matching charts. Ó 1999 Carl G. Lodström.

X_C = 50 Ohm at 860 MHz? X_C = 1/(2 x p x 860 x 10⁶ x 50) = 3.7 pF. Very likely! You could also get rid of this by adding a small inductor in series, then just match the real parts (100 Ohm -> 50 Ohm). How to calculate the inductance? L = 1/((2 x p x 860 x 10⁶)² x 3.7 x 10^-12) = 9.3 nH. A little 1-turn loop of component wire on a pencil is my guess! For input matching: turn the figure around so the 'Z' goes tot the transistor and 50 Ohm to the antenna! Let's do an input match example!

A small bipolar transistor may have R_IN = 10 Ohm and an inductance of 3 nH.

3 nH at 860 MHz, X_L = L x 2 x p x 860 x 10⁶ = 3 x 10^-9 x 2 x p x 860 x 10⁶ = +j16.2 Ohm, it should be matched with -j16 Ohm. What capacitance will resonate with 3 nH at 860 MHz? C = 1/((2 x p x 860 x 10⁶)² x 3x 10^-9) = 11.4 pF.

So, looking in the lower diagram, there may be solutions for this at a very short L₁ and for L₁ about 4". The shorter is more attractive! At about L₁ = 0.2 ~ 0.3" there ought to be a -j16 with -j12 just to the right. From the upper graph, in the same place, we see the 12.5 Ohm line disappearing at the left around 1.3" up! Looking in the lower graph again 1.3" up at the left edge, yes! That is close to -j16. So L₁ of no length and L₂ of ~ 1.3" should do it! This open ended stub direct to the transistor input and feed 50 Ohm in at the same point!

Remember: for each L₁ - L₂ point in one graph, you have to consider the same L₁ - L₂ point in the other. They are to be thought of as overlaid on each other, with the open tip of L₂ as the "pointer." Notice how the lines are perpendicular everywhere!

The other day I was finally going to throw out that box with old computer junk! The 5¼" drives with 360 kB capacity are no longer hot stuff. As I lifted the drive for a trip to its last resting place, I saw that flat motor. Hmm, a big, fat, IC, cable with many leads. It got to be a mess to drive that motor. Or is it? A closer look revealed that the 4-lead cable surely had one Ground and one '+', leaving two to go. Another one marked 'C' and the fourth one marked '5'. With a little bit of detective work, not even hard, I found out the hookup:

The 'C' survived 12 V but should probably have 5 V. It has a 10k resistor in series.

The circuit proved to have a few interesting properties. It is a 3~ motor with feedback, probably from Hall (magnetic)-sensors. Once it is up to speed it is phase locked to a frequency reference! The little green rectangle marked Y101 is a 800 kHz resonator. A crystal will probably work, but I have not tried yet. The motor frequency is 50.0 Hz for a speed of 5 r/s or 300 rpm. So there is an internal divider of 16,000:1 in the IC. Loading the motor mechanically with up to at least 50 mNm had no effect on the speed! As I loaded the motor, the voltage on the windings increased in amplitude!

The current consumed from 12V is 62 mA, with no load on the motor. It increases with load up to 700 mA for a stalled motor. Loading resulting in a permanent current of more than 300 mA should probably be avoided since the IC gets pretty hot. The resistor marked R106 (0.51W ) is the current return to ground. Across this resistor is a good place to measure the current. Increasing the value of R106 will probably lower the current limit.

The voltages to each phase looks like a stair case with 5 levels. A "simulated" sine wave. If one need 3-phase for something, this wave form is easily filtered to a sine wave. It can be amplified and drive some big bad motors... Next, of course, I have to find out if one cannot drive it from an external frequency source, and if so, what frequency range can be used. As I loaded the motor, the voltage on the windings increased in amplitude!

I wound a string around the output shaft (33 mm diameter), connected it to a dynamometer (calibrated in gram, so it is a scale.. One 'g' = 9.81 mN) and measured the current consumption for various loads:

If I figured it right, 1 oz-in = 7.0765 mNm (milli-Newton-meter).

The spindle is exceptionally well centered and probably supported by double ball bearings. it is strong, silent and have no play that I can notice. It appears perfectly well centered. A very nice spindle.

The other motor is a stepper motor that used to move the heads radially. It has a 21.5 mm stroke. I have not yet measured how many steps, but maybe 50 or so. This motor appears to be the 'unipolar' type for which nice translators ("driver IC") exist. Allegro is a company that have several nice ones. The IC takes care of driving the windings, given just pulses in.

I have not figured out what to do with this motor!

Maybe a little lens grinding and polishing machine? The stepper motor can push a grinding pad back and forth on the rotating lens… Download OSLO 4.5 and design your own eyepieces for the telescope from scratch!

Maybe a meter for viscosity? If the rotating shaft is attached to a rotating cone with a large angle, 160° or so, lowered tip down in a liquid so the tip almost touch the flat bottom of the vessel, the current consumption will reflect the viscosity of the liquid. or gas.

Maybe a rotating table for the clay artist? The clay can be smacked right onto the motor! It has a 74 mm (~3") large , almost flat, rotating part.

Maybe it can go at 400 Hz with a 6.4 MHz crystal. I can have use for that. Some gyroscopes, servo motors and syngons/elgons use 400 Hz/3~. The three phases are available at the soldered wires marked W102, W103 and W104. With three power amplifiers (National Semiconductors' LM1875 for 3 x 20W or LM3875 for 3 x 56W), some filtering, a 3~ transformer and there is the nice 115V/400 Hz/3~!

We live in a fantastic era when anybody can afford a very good time keeper! For US$ 10 ~ 20 you can buy a wall clock or an alarm clock with a quartz movement. They usually run better than to a minute/month or 1 part in 43,200. If you check them against a time signal or some radio program you will find that the error in the clock is quite accurate! The clock may gain 17 seconds/month, one month after the other! This indicates that something can be done about it! Radio programs may be tricky to set the clock after. The announcer may have a digital clock, not showing seconds, so when he say it is 08:59 it may be 08:59:58 and 09:00:02 before he finishes the sentence.

I have noticed that "Morning Edition" and "All things considered" on the Public Radio start on the second.

Then you need a frequency counter (or you can make it a long-term project adjusting a little every month and keeping a log) with ability to measure period time. Most frequency meters are not very accurate in themselves, but can be adjusted. For this job it is optional. Either you know how by much your clock is off and you just use the stability over an hour of the counter to trim the clock or you have an accurate counter and can take on the clock not knowing how much it gains. They usually gain! We will see later why.

The Clock:

Historically, the first ones, from the early 70's, had a 4,194,304 Hz oscillator. 2²² Hz. By dividing the frequency by 2 22 times one get 1 Hz. These clocks had excellent stability and usually a trimmer capacitor as well. It consumes power (but not much, a C-cell lasted a year) to do all these divisions, so for watches it was not the way to go. Very small 2¹⁶ (65.536 Hz) and 2¹⁵ (32,768 Hz) crystals were developed, but their temperature stability left a lot to be desired. This has been remedied since at least 15 years and now very small 32 kHz crystals, with excellent temperature stability, rule!

There are two kinds of clocks, analog and LCD. In neither of them should you measure direct on the crystal, it pulls the frequency, making for an useless measurement. It may be helpful to have an oscilloscope as a "pre-amplifier" so you can see what you measure on. Connect the counter to "Y2 OUT" or what output you have from the scope. If none, connect it in parallel with the scope input.

On the LCD you can find the back plane frequency divided down from the crystal, usually 32 or 64 Hz. Just probe all leads, pads, test points and what you can reach until you find a nice square wave!

In the analog clock you may have to measure the voltage across the coil of the motor, detecting the 1 s pulses. In some clocks the pulse is reversed every other second, so the counter will make one measurement every two seconds. It is possible that you can find a divided down frequency for the alarm (if any). Poke around!

To trim the counter:

You need a shortwave receiver that can listen to WWV, WWH or some other frequency standard stations. Turn on the counter and let it warm up for an hour! Place it near the receiver or the antenna and hear a tone interfering with the standard station. As the counter warms up, the tone will stabilize. There is a small trimmer capacitor somewhere near the crystal, adjust it for zero beat. The counter is now very accurate!

Adjusting the clock on a trimmed counter.

If you know that the clock is fast by, say, 17 s/month as in the example above. It is an analog clock and you have found the 2s signal and trigger the counter on each cycle. With a 10 MHz reference in the counter you will read 1,999,986.9 µs Adjust your clock is for a 2,000,000.0 µs reading and it is right on!

If the clock is an LCD and the back plane frequency is 32 Hz: Measure the period time for 100 cycles! It will take about 3 s per reading. 1/32 s = 31250 µs, but 100 cycles of it, and a 10 MHz clock rate in the counter will give you a reading of 3,125,000.0 µs. Quite accurately, one count in the last digit equals one second per year! If the clock is 17 s/mo fast you will get a read of 3,124,979.5 or 20.5 µs short of 3.125 million.

Not trimmed the counter, but know how the clock is running:

With an analog clock your counter will show something in the neighborhood 20 million µs. Adjust the clock for a reading that is 131 µs larger.

With a LCD clock (and 32 Hz frequency) your counter will show something in the neighborhood of 3.125 million µs. Adjust the clock for a reading that is 20.5 µs larger.

If you are sure about your clock error, and have corrected it, now you can adjust the counter for a reading of 2,000,000.0 or 3,125,000.0 if it is off by more than 1 part per million (20 or 31 counts).

Sounds great, but how to trim the clock?!

You too may have noticed how these quartz clocks gain time if they are not right on. The crystals are made for to have a small trimmer capacitor across it, but the manufacturer saved a few cents by not putting in a trimmer capacitors and a few more cents by not having somebody trimming it. You can put trim back in the clock!

The little aluminum cylinder here is the crystal. Next to it is a trimmer capacitor that happen to fit in this particular clock. On the right is a way to make a trimmer capacitor out of enameled wire when there is not much room. The longer it is, the more capacitance. Start with one longer than needed, loosely twisted! Cut it until just past the right value, then twisting the wires tighter until you are back on the right value! I used this version for a Sony ICF - 7600DS. Fancy name, but they could not put a trimmer on the LCD clock! It is three months since I set it the clock is still within 1 second of correct time. This is better than I had hoped for.

First you have to get to know your Ohm meter! With an other meter, find out if the Ohm meter puts out + on its + or - terminal! Many, but not all puts out + on the - terminal. It makes for a simpler internal range switch. While at it you may also measure the open circuit voltage and the short circuit current from your Ohm meter. Try it on all the ranges.

For the transistor checking, a simple analog meter, or an old Vacuum Tube Volt Meter (VTVM) is the preferred instrument. The VTVM uses a 1.5 V battery on all ranges and the analog meter usually does the same but for a high range, like R x 10k or more.

The internal resistance in the instrument limits the current. Look for example at the instrument dial way up on this page, where a RF power meter is described. The outermost scale is the Ohm meter. You can see that it's midvalue is 10. This means that on the R x 100 range it has an internal resistance of 10 x 100 = 1000 Ohm. The battery is 1.5 V, so the short circuit current is 1.5/1000 = 1.5 mA. For a resistance leaving the needle midway the current is of course 0.75 mA and the remaining voltage 0.75 V.

A DMM is not so useful, but if it is all you have it can do some work. As an Ohm meter it is often a "2.000 V voltmeter" measuring the drop across the load for a constant current. At the 2k range the current is often 1.000 mA, at the 20k range it is 100.0 µA and so on. At the 200 Ohm range the current is often still 1.000 mA, but the voltmeter is a 0.2V voltmeter. The current usually comes out of the - terminal.

Now to the transistors. You have to bear with the quality of the sketches. I will use whatever tools Word for Windows allows since I do not have time now to do them nicely. Maybe I will replace them later.

Assume that you don not know anything about the transistor! Not if it is a J-FET, a MOS-FET or a bipolar. Not even if it is Silicon or Germanium device. It may even be a thyristor (SCR). All you know is that it has three leads!

Assuming for now it is a bipolar transistor, pin 2 must be the base. Connect the meter to 1 & 3. Wet a fingertip and put it between 2 and the other pins, one at a time. If nothing happens, reverse the meter leads and try again. When you get a deflection, much more than when you put the same wet finger direct between the meter leads, you have found the collector. The remaining pin is the emitter. Of course, you may also have a faulty device, and this is probably the case if you do not find the behavior to fit any case described here.

If you find a diode in one direction and the third pin seems "open circuited" no matter polarity, you may have a MOSFET. These are usually power devices, residing in TO-3 or TO-220 packages. They can also be modulated with static electricity but have a very high capacitance on the gate (pin 2) often 1000 pF and sometimes as much as 10 nF. The Source and Drain are not interchangeable due to the internal parasitic diode (that can handle as much current as the transistor itself).

I have to leave it at this for the moment, but may come back with SCR's and some other devices.

A DESIGN ERROR IN THE IC-R7000 RECEIVER.

Consider the schematic diagram below:

The Operational Amplifier is one half of an industry standard more commonly known as LM358. A dual amplifier with PNP inputs allowing for operation on a single supply with the inputs on ground potential.

This is very useful in general, but not needed here since the supplies are +9 and -7 V. The circuit is in series with the audio output (loudspeaker and Recorder output). It is intended to notch out (remove) a 5 kHz tone that may come from the synthesizer. As all other OP amplifiers, the inputs of the -358 have a "Bias Current". That is: a very weak current wants to flow out of the inputs. If I remember right, this is on the order of 50 nA for the -358. For this reason there must be a DC path from both inputs to something "solid." The - input goes to the output, that is OK. Look at the + input and imagine that you are a DC current trying to get out!

There is no DC path! Capacitors everywhere! So the + input will drift up (and the output will follow) to the +9V rail. The amplifier reaches the limit of its linear range and the sound gets distorted. Turn the radio off and on again, and it sounds well! When the supply voltages collapse the capacitors are "reset" to 0 Volt. If they or the circuit board leaks a little the problem may never show, but it is there, waiting for a dry day to appear!

A few MW from anywhere in the RC network to ground will fix the problem.

4.7 MW from the minus side of C135 for example.

I have recently moved and have not found my solder iron yet so I rigged a resistor between a ground shield to the - side (the can) of C135 and it works fine. A little Vaseline may fend off oxidation. This is not a recommended solution. The resistor may come loose and will of course land where it will do the most damage, so at least one end should be soldered. The small movements from when the radio is turned on and off (warming up - cooling off) are probably sufficient for making a lasting contact if the ground end is soldered in place. The + input of the -358 is anther place to solder the resistor to.

Now and then other little things will appear on this page. Click in again!

Drop me a line!